asked on

Configuring FallBack for a Solaris Machine.

Hello All,

Have been given the (un)enviable task of doing a feasibility study for configuring a fallback server for a Solaris machine, when the situation is that I have no idea what I am supposed to do.

Can any of you gentlemen/ladies direct me to a resource where I can get a crash course / tutorial about this topic? Something that will explain to me what fallback is, how it works etc etc.

Thanks and regards,
Sourish.

SOLUTION

gheist

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

ASKER CERTIFIED SOLUTION

tfewster

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

colinbo

The term "fallback" seems vague at best. What is the objective? If your are looking to implement an upgrade to the OS and need to be able to backout of the change if it all goes pearshaped, then you could try the Live Upgrade feature of Solaris 9, where, if you have mirrored OS volumes, you detach one submirror (assuming DiskSuite and not VxVM is being used), then apply the OS upgrade to the offline submirror then schedule a controlled reboot from the offline submirror disk. If applications or other things fail to start, you just reboot again from the original disk (which has remained unchanged) and you're back and running again.

More details about Live Upgrade can be found at:
http://wwws.sun.com/software/solaris/liveupgrade/
and
http://www.sun.com/bigadmin/features/articles/live_upgrade.html

- CB

xaverian_dgp

ASKER

Well, further enquiries into the requirement gave rise to the fact that FALLBACK was almost synonymously being used with HIGH AVAILABILITY. Is that okay. I guess i am looking for HA options.
Thanks.

Nukfror

As previously suggested, there is Sun Cluster but there is also Veritas Cluster Server.

Live Upgrade provides an option but from what I remember, its kinda a PITA on systems with SVM/VxVM enabled. LU doesn't support mirrored root disks from what I recall. But colinbo may have had better experiences with this then I have.

Complete poor man's fallback is you setup SVM to mirror the internal disk, you break the mirror and remove the mirror disk. Put that disk into a same type machine e.g. v440 boot disk into another v440. Boot the second machine off a CDROM and remove mirror elements from the disk (/etc/system tweaks and /etc/vfstab tweaks). You could also leave the "mirrored" disk in the first server until needed and have a script do the resync, mirror breaking, and file tweaks nightly or whatever.

I've seen one major TELCO do just the above.